I have implemented numerous Advanced Capture solutions over the years and have identified a number of sticking points that are easily overcome, but can also be a real pain. I would classify an Advanced Capture solution as any document capture implementation that is more than a standard heads-up indexing or very basic zonal OCR of fixed forms based on X and Y coordinates. That definition leaves a lot of room for what could be designated as an ‘advanced capture solution’, but I think it fits. Once you move out of the realm of basic capture, you start to encounter a lot of the same problems. There is one problem I seem to encounter every time we implement a solution.
Not enough sample pages when configuring the solution
I can’t tell you how many times I have been engaged to implement a project and only given one or two sample pages. It seems crazy to me that someone could think that there might be a positive result from this approach, but believe me, it still happens all the time. If you trudge along and configure the solution with your minimal sample set, you will enjoy a very long and drawn out beta testing and implementation. There is always variation when talking about documents, even fixed form documents. Whether it is different printers, different content creators, or just different versions of the same form, there is always variation. Variation has to be accounted for, or else the solution won’t work, that is why configuring a solution with only one or two sample pages will cause a long implementation. Every time you encounter a variation, you will inevitably have to go back and reconfigure the solution. Some of this can be mitigated with how you configure the solution, but having sufficient samples is still a must. Even if using a software package with ‘learning’ capability, it is best to get a lot of samples up front.
How many samples you ask? The answer is seems pretty simple, to configure an advanced capture solution you need as many samples as possible to cover every possible variation of a form. If you get that many samples and configure the solution to work with and account for all variations, you end up with a much more bullet proof configuration come beta testing and implementation time. A good rule of thumb would be to have a minimum of 10 samples for every document type, but ask for 50 and reserve the right to ask for more. 50 might seem like a lot but in the scheme of things it is nothing. A company wouldn’t implement an advanced capture solution unless they could justify the steep price of it, the only way to justify that price it implement these solutions for processes that process thousands (if not tens of thousands) of documents a day. 50 sample documents for a solution that is expected to handle thousands of documents a day is not much and you’ll hit the ground running when beta testing starts rather than grinding to a screeching halt.